An automatic search and alignment method for crystal's self-alignment based on machine learning on high power laser facility

The process of online crystal alignment in high-power laser facilities is both challenging and labor-intensive. We propose an automated method for self-alignment of crystals on these facilities, utilizing machine learning. This method employs a machine learning algorithm running on a Raspberry Pi to automatically locate the reflective spot from the crystal's back surface and adjust its position to achieve self-alignment. The proposed scheme comprises two modules: a rectangular spiral spot scanning search method module and an automatic spot aligning method module based on the open-source machine learning algorithm M-LOOP. Initially, the rectangular spiral spot scanning search method is used to position the laser spot within the field of view of the CCD camera. The M-LOOP algorithm then automatically adjusts the laser spot to align with the reference center. The hardware system includes a crystal alignment optical system, motors, a CCD camera, and a Raspberry Pi. Offline experiments have demonstrated that this method can accomplish the automatic search and alignment of the crystal return spot of a He-Ne laser in approximately 10 minutes. This solution addresses the issue of traditional technology requiring manual search and adjustment of the crystal's reflection spot for self-alignment.